1. COSIRES2004
MD Simulation of Surface Smoothing due to Cluster Impact: Estimation of Radiation Damage
T.Muramoto, K.Itabasi and Y.Yamamura
Okayama University of Science, Department of Informatics
Ridai-cho 1-1, Okayama , Japan
The radiation damage of irradiated surfaces by cluster ion with a few eV/atom is studied through MD simulations, where (Ar)3055 clusters with 3-10eV/atom are bombarded on a rough Cu surface.

2. Collision cascade
100fs: Channeling is found.
150fs: Dechanneling is caused.
250fs: Most of cascade is formed.
400fs: Focusing is appeared.
Finally, Frenkel defect is induced by monatomic ion beam.
Using DYACOCT code dynamical simulation of atomic collision in crystal target based on binary-collision approximation

3. Feature of Cluster ion beam
Chemical-mechanical polishing
Standard smoothing technology. Wet process using chemical abrasive and grinding pad. Unsuitable for soft material. Surface cleaning is essential to the contamination.
Ionized cluster beam
Low charge and velocity. High energy and particle density. Beam intensity consistent hardly with control of cluster size.
Low velocity and charge to mass ratio leads to low damage.
High particle density formation.
Explosion to lateral direction.
High deposited energy density results in high-yield sputtering.

4. Measurement of fractal surface
Monolayer mole numbers on porous silica gel as a function of molecule cross-section.
Surface may be rough and even fractal down to the molecular size range.
D=3.02±0.06
[P.Pfeifer, D.Avnir and D.Farin, J. Stat. Phys. 36 (1984) 699.]

5. Fractal Similarity dimension D=logA/logB: A parts, scaled by ratio 1/B
Shape and phenomena with non-characteristic scale
Fractal is characterized by dimension with non-integer value.
Similarity dimension D=logA/logB: A parts, scaled by ratio 1/B
Line
2 parts, scaled by ratio 1/2
The von Koch snowflake curve
4 parts, scaled by ratio 1/3
4 parts, scaled by ratio 1/2
Plane

6. Contents 1. MD simulation model 2. Surface smoothing
1.1. Projectile and target information, interatomic potentials
1.2. Control of target temperature
1.3. Initial rough surface and fractal
2. Surface smoothing
3. Radiation damage
3.1. Criterion of damage type
3.2. Quantification of damage
3.3. Thickness of damage layer

7. Interaction potential
MD simulation model
Interaction potential
Long-range
Short-range
Ar-Ar
Lennard-Jones
AMLJ
Cu-Ar
Cu-Cu
EAM
Cluster energy: 3, 6.5, 10eV/atom
Periodic boundary:
Thickness of target: 11, 22nm
LMD layer (300K): 1.0nm
Cycle of impact: 20ps
Roughness: 1.5nm
Fractal dimension: 2.5

8. Control of target temperature
Cool down
by LMD layer
Total kinetic energy
of target atoms
Cluster energy
Numerical cool down
300K
1st
2nd
Time of simulated system

9. Shock wave in view of temperature
This is a color map of temperature in cross-section. It found that the shock wave reflect at bottom. Average sputter yields are 58 and 54, respectively. This difference is not a significant error.

10. Initial fractal surface
Fourier Filtering Method

11. Experiment of surface smoothing
Experiment by Kyoto university’s group:
H. Kitani et al., Nucl. Instrm. Meth. B121 (1997) 489.
Energy: 20keV/cluster
Dose: 50 ion/nm2
Initial roughness: 5.9nm
Final roughness: 1.0nm

12. 3eV/atom (Ar)3055
Sputter yield per impact = 0.0, RMS Roughness = 0.3nm
Surface shape changes slowly.

13. 10eV/atom (Ar)3055
Sputter yield per impact = 54, RMS Roughness = nm
Surface shape changes rapidly.

14. 6.5eV/atom (Ar)3055
Sputter yield per impact = 5.9, RMS Roughness = nm
Experiment: 50 ion/nm2, 20keV Ar3000 bombardment on Cu
[H. Kitani et al., Nucl. Instrm. Meth. B121 (1997) 489.]

15. Development of average roughness
Final roughness is determined by
the magnitude of surface modification by individual cluster impact.
Fractal surface satisfies the scaling relation of self-affine:
Z(ax,ay)=a3-DZ(x,y).
[J.Feder, FRACTALS, Plenum, New York, 1985.]

16. Radial distribution: 10eV/atom
The irradiated targets are cooled down to 3K in 50ps using LMD method for the whole target atoms.

17. Bond angle distribution: 10eV/atom

18. Stacking fault
Rough estimation of Pressure:
17GPa (3eV/atom)
36GPa (6.5eV/atom)
55GPa (10eV/atom)
Shear modulus: 48GPa
Plastic deformation with high pressure at impact produced stacking fault.

19. Local crystal direction
Local x, y, z axis are determined from the position of first neighbor atoms.
Angle between Local and global x, y, z axis is written as qx, qy, qz, respectively.

20. Angle between local and global axis

21. Crystal grain
Melting and crystallization of impact region produced grain.

22. Number of first neighbors = 11
There are cage and linear groups near the impact region.

23. Vacancy
This is a quenched-in vacancy when target cooled down to 300K.

24. Divacancy
This is a quenched-in vacancy when target cooled down to 300K.

25. Vacancy cluster
This is a quenched-in vacancy when target cooled down to 300K.

26. Dislocation
Number of first neighbors is 13 (red) and 11 (blue). Green symbol is stacking fault atoms.
ABCA B C A
CABCA
This is an edge dislocation. The tensile and compressive stress acts on the red and blue atom, respectively.

27. Correlation between radius and size

28. Number of displacement atoms: after 16th impact

29. Number of displacement atoms: 10eV/atom

30. Thickness of damage layer
Thickness are about 8nm (10eV/atom) and 4nm (3eV/atom).

31. Summary
The average roughness due to 6.5eV/atom cluster bombardments ranges about nm, which is less than the result of experiment* from the smallness of target in this simulation.
There is no Frenkel pair, because the cluster impact with a few eV/atom cannot produce the energetic PKA.
In the impact region, the high temperature results in a few vacancy and grain, and the high pressure generates some dislocation and stacking fault.
The pressing effect is important than the thermal effect to induce the damage in the big cluster impact with a few eV/atom.
The big cluster ion can affects only near the surface, which the thickness of damage layer is about 4-8nm for 3-10eV/atom.
* H. Kitani, N. Toyoda, J. Matsuo and I. Yamada , Nucl. Instrm. Meth. B121 (1997) 489.

32. Acknowledgement This work was supported by a Grant of The Academic
Frontier Project promoted by The Ministry of Education,
Culture, Sports, Science and Technology of Japan.